Project Summary The proposal ?X-ray data analysis in the presence of structural variability? aims to advance diffraction data analysis methods so that the variability between crystals and within crystals is optimally modeled during data processing in reciprocal space and during structural analysis in real space. The significance of the proposed work results from the importance of the technique, which generates uniquely-detailed information. X-ray structures are used to understand cellular processes at the atomic level directly, to explain and validate results obtain by other techniques, to generate hypotheses for detailed studies of cellular process, and to guide drug design studies ? all of which are highly relevant to the NIH mission. Macromolecular crystals are frequently of limited size and crystal lattice order. Both may result in the need for combining data from multiple crystals for successful structure solution, with the limited order generating diffraction artifacts and correlating with non-isomorphism between different specimens. Non-isomorphism hinders the averaging of data sets from multiple crystals, because for successful averaging, data need to be very similar. The problems with averaging are compounded by incompleteness of the data in a single data set, radiation-induced changes in the crystal under investigation, and lack of statistical measures that would inform experimenters regarding whether or not the data analysis is progressing in the right direction. There are also technical challenges associated with averaging multiple data sets that result from the combinatorial complexity of data analysis when a large number of data sets need to be analyzed. Final difficulty appears when the analysis of the structural results obtained from multi-crystal experiments must separate the desired biological signals, e.g. the presence of a ligand or a specific dynamic behavior of the molecules, from the noise. Our proposal addresses these problems by developing and implementing innovative approaches. In Aim 1, new approaches will be developed and implemented for averaging multiple, potentially incomplete data sets resulting from one or more crystals. Owing to our innovative approach to modeling the components of non- isomorphism, we expect that even quite non-isomorphous data sets can be used together to solve challenging structures. In Aim 2, methods that will analyze the results of averaging data sets from multiple crystals in real space will be developed. The descriptors of averaging will be correlated with the outcomes of the structural analysis, so that the contributors to variability in real space can be quantified and interpreted. Finally, in Aim 3, a web-based server will be developed in order to provide these methods to the structural biology community.